Complexity increases with number of treatment groups. Quadratic increase between the number of treatment groups (X axis) and the number of pairwise comparisons (right Y axis) is shown in black. The exponential increase in the number of post hoc patterns (left Y axis) is shown in gray. Note that by the time there are seven treatment groups, we predict nearly a billion different patterns. However, due to the lack of independence among the comparisons, the actual number of possible comparisons generated in 100 iterations of a 10,000 gene, 30 array model data system (dotted gray line) is considerably less.
Hulshizer and Blalock BMC Bioinformatics 2007 8:240 doi:10.1186/1471-2105-8-240